Evaluation of an autoregressive spectral estimator for modal wave number estimation in range-dependent shallow water waveguides

In shallow water, geoacoustic properties of the seabed can be inferred from knowledge of acoustic normal modes propagating in the waveguide. For range-varying waveguides, modal content varies locally in response to changes in the environment. The problem becomes estimating the local modal content of a propagating field. For acoustic fields measured on horizontal arrays, resolution of closely spaced modal eigenvalues depends on the data aperture length. Therefore, range variability in a waveguide can only be detected for range scales of order of the resolvability of the individual modes. When only short data apertures are available for modal estimation, high-resolution methods must be used. In this paper, a high-resolution autoregressive (AR) spectral estimation method for extracting the modal information from measurements of a continuous wave acoustic field made on a horizontal array is evaluated. Performance is discussed for estimation accuracy and resolution in the presence of noise. Results are compared to previous work characterizing high-resolution wave number estimators. The AR estimator performance is comparable to other high-resolution methods and does not require prior information about the number of propagating modes. In addition, range/wave number plots obtained using a sliding window technique yield a robust method for identifying propagating modes and any changes with range. (c) 2006 Acoustical Society of America.